报告时间：2024年9月18日9:30-11:30

报告地点：必赢线路检测中心新大楼925会议室

报告人：徐贯东教授

工作单位：香港教育大学

举办单位：必赢线路检测中心

报告简介：

With the exponential growth of information, Recommendation Systems (RecSys) have become pivotal tools in managing data overload. Both in the academic and industrial spheres, the use of graphs for structuring recommender systems data and applying graph neural network technology for prediction have become areas of intense research focus. However, graph neural networks bring opacity issue to recommender systems, which leads to unexplainable recommendation results and biased learning processes and results. To address these two key research challenges, our studies have focused on the explainability and fairness of recommender systems. Specifically, for the explainability of recommendations, we studied user intent disentanglement, path exploration in graphs, temporal modeling of paths, and counterfactual learning for reasoning. For the fairness of recommendations, we studied selection bias in static scenarios and bias problems in dynamic scenarios. These research efforts have led to satisfactory results in recommendation outcomes, explainability, and fairness. They have also made meaningful theoretical explorations and experimental innovations for building reliable and credible recommendation systems in the future.

报告人简介：

Guandong Xu is a Chair Professor of Artificial Intelligence at Education University of Hong Kong (EdUHK). Before joining EdUHK, he is a full Professor in Data Science at the School of Computer Science and Data Science Institute, University of Technology Sydney, with PhD degree in Computer Science. His research interests cover Data Science, Recommender Systems, User Modelling, and Social Computing. He has published three monographs in Springer and CRC Press, and 220+ journal and conference papers, including TOIS, TKDD, TKDE, TNNLS, TCYB, TMM, TSE, TSC, TIFS, VLDB, IJCAI, AAAI, SIGMOD, KDD, SIGIR, CVPR, NIPS, WWW, WSDM, ICDM, ICDE, ICSE, and FSE conferences. He is the Editor-in-Chief of Human-centric Intelligent Systems and the assistant Editor-in-Chief of World Wide Web Journal. He has been serving on the editorial board or as a guest editor for several international journals, such as TOIS, TII, TCSS, PR etc. He has received several Awards from the academic and industry community. He is elevated as a Fellow of the Institute of Engineering and Technology (IET) and Australian Computer Society (ACS) in 2022 and 2021, respectively.